What happened.
Pangram’s five-platform study found that one in four longer social-media posts it reviewed were entirely AI-generated. LinkedIn stood out in the findings: 41% of its long-form posts were flagged, the highest share described in the research record.
LinkedIn also made up roughly one-third of the posts scanned, yet accounted for nearly two-thirds of all AI content detected in the study. That concentration is the central finding behind the claim that LinkedIn leads in long-form AI “slop.”
Why it matters.
The figures add a concrete measure to a broader trust problem around AI-generated and AI-assisted output. When readers see professional-looking long-form posts, it may be harder to tell whether the apparent human voice, experience, or analysis is actually human.
That uncertainty matters especially where social posts are used to build credibility, share expertise, or shape professional discussion. The study does not establish why LinkedIn had the highest detected share, but it indicates that the platform deserves closer scrutiny in discussions about the reliability of online text.
The limit and next receipt.
The evidence is limited to Pangram’s study and a light, single-source news record, so confidence is medium. Pangram’s detection is described as conservative, which means the reported figures may be an undercount; it also means the record does not provide a complete measure of all AI-written content across the platforms.
The next useful receipt is the study’s underlying methodology and any independent replication using comparable long-form posts across the same platforms. That would help show whether the reported LinkedIn gap holds beyond this scan.
Watch for Pangram’s methodology and an independent replication of the five-platform results.
Upstream references
Digest dated 2026-07-13 · upstream model claude-sonnet-4-6. Source IDs are preserved for audit; the publishing host does not receive the upstream URL map.
- 1
20f16269d1dea5a07d21fee530eb41491dbc8af1Reference from the upstream research server
This quick brief was generated by Terra from a dated upstream research digest. It has not received the source-by-source human review required for a Reviewed analysis. Material limit: The record provides a single study with medium confidence and no full methodology; conservative detection may undercount AI-generated content.